|Scour Technology | Bridge Hydraulics | Culvert Hydraulics | Highway Drainage | Hydrology | Environmental Hydraulics|
|FHWA > Engineering > Hydraulics > HEC 23 v2 > Design Guideline 11 Rock Riprap at Bridge Piers|
Bridge Scour and Stream Instability Countermeasures: Experience, Selection, and Design Guidance-Third Edition
Design Guideline 11 Rock Riprap at Bridge Piers
When properly designed and used for erosion protection, riprap has an advantage over rigid structures because it is flexible when under attack by river currents, it can remain functional even if some individual stones may be lost, and it can be repaired relatively easily. Properly constructed riprap can provide long-term protection if it is inspected and maintained on a periodic basis as well as after flood events. This design guideline considers the application of riprap as a pier scour countermeasure.
Design of a pier scour countermeasure system using riprap requires knowledge of: river bed and foundation material; flow conditions including velocity, depth and orientation; pier size, shape, and skew with respect to flow direction; riprap characteristics of size, density, durability, and availability; and the type of interface material between the riprap and underlying foundation. The system typically includes a filter layer, either a geotextile fabric or a filter of sand and/or gravel, specifically selected for compatibility with the subsoil. The filter allows infiltration and exfiltration to occur while providing particle retention.
Bridge pier riprap design is based, primarily, on research conducted under laboratory conditions with little field verification. Flow turbulence and velocities around a pier are of sufficient magnitude that large rocks move over time. Bridges have been lost due to the removal of riprap at piers resulting from turbulence and high velocity flow. Usually this does not happen during one storm, but is the result of the cumulative effect of a sequence of high flows. Therefore, if rock riprap is placed as scour protection around a pier, the bridge should be monitored and inspected during and after each high flow event to insure that the riprap is stable.
The guidance provided in this document for pier protection applications of riprap has been developed primarily from the results of NCHRP Project 24-07(2) (Lagasse et al. 2007), NCHRP Project 24-23 (Lagasse et al. 2006), and NCHRP Project 24-07(1) (Parker et al. 1998).
The basic mechanism causing local scour at piers is the formation of vortices (known as the horseshoe vortex) at their base (Figure 11.1). The horseshoe vortex results from the pileup of water on the upstream surface of the obstruction and subsequent acceleration of the flow around the nose of the pier or abutment. The action of the vortex removes bed material from around the base of the obstruction. The transport rate of sediment away from the base region is greater than the transport rate into the region, and, consequently, a scour hole develops. As the depth of scour increases, the strength of the horseshoe vortex is reduced, thereby reducing the transport rate from the base region. Eventually, for live-bed local scour, equilibrium is reestablished between bed material inflow and outflow and scouring ceases. For clear-water scour, scouring ceases when the shear stress caused by the horseshoe vortex equals the critical shear stress of the sediment particles at the bottom of the scour hole (Richardson and Davis 2001).
In addition to the horseshoe vortex around the base of a pier, there are vertical vortices downstream of the pier called the wake vortex (Figure 11.1). Both the horseshoe and wake vortices remove material from the pier base region. However, the intensity of wake vortices diminishes rapidly as the distance downstream of the pier increases. Therefore, immediately downstream of a long pier there is often deposition of material.
Factors which affect the magnitude of local scour depth at bridge piers are (1) velocity of the approach flow, (2) depth of flow, (3) width of the pier, (4) length of the pier if skewed to flow, (5) size and gradation of bed material, (6) angle of attack of the approach flow to the pier, (7) shape of the pier, (8) bed configuration, and (9) ice formation or jams and debris.
Most of the early work on the stability of pier riprap considers the size of the riprap stones and their ability to withstand high approach velocities and buoyant forces. Secondary currents induced by bridge piers cause high local boundary shear stresses, high local seepage gradients, and sediment erosion from the streambed surrounding the pier. The addition of riprap also changes the boundary stresses.
There are at least a dozen equations for sizing bridge pier riprap that can be considered for design (Lagasse et al. 2007, Melville and Coleman 2000). Typically, the stability of riprap is expressed in terms of the Stability Number, Nsc which is used in numerous equations to size riprap. This approach, which derives from the work of Isbash (1936) considers turbulence intensity to determine rock size (see Figure 11.2). Riprap stone size is designed using the critical velocity near the boundary where the riprap is placed. However, many of the pier riprap sizing equations are modified versions of bank or channel protection equations and, therefore, the use of this approach has limitations when applied at bridge piers because of the strongly turbulent flows near the base of a pier. Most of the remaining equations are based on threshold of motion criteria or empirical results of small-scale laboratory studies conducted under clear-water conditions with steady uniform flow.
To determine the required size of stone for riprap at bridge piers, NCHRP Project 24-23 recommends using the rearranged Isbash equation from the Federal Highway Administration's Hydraulic Engineering Circular No. 23 (Second Edition) (Lagasse et al. 2001) to solve for the median stone diameter:
It is important that the velocity used in Equation 11.1 is representative of conditions in the immediate vicinity of the bridge pier including the constriction caused by the bridge. If the cross-section or channel average velocity, Vavg, is used, then it must be multiplied by factors that are a function of the shape of the pier and its location in the channel:
If a velocity is distribution available from stream tube or flow distribution output of a 1-D model or directly from a 2-D model, then only the pier shape coefficient should be used. The maximum velocity in the active channel Vmax is often used since the channel could shift and the highest velocity could impact any pier.
Once a design size is established, a standard gradation class can be selected, if design criteria and economic considerations permit. Using standard sizes the appropriate gradation can be achieved by selecting the next size larger size class, thereby creating a slightly over-designed riprap installation, but economically a less expensive one.
Riprap design methods typically yield a required size of stone that will result in stable performance under the design loadings. Because stone is produced and delivered in a range of sizes and shapes, the required size of stone is often stated in terms of a minimum allowable representative size. For pier scour protection, the designer specifies a minimum allowable d50 for the rock comprising the riprap, thus indicating the size for which 50% (by weight) of the particles are smaller. Stone sizes can also be specified in terms of weight (e.g., W50) using an accepted relationship between size and volume, and the known (or assumed) density of the particle.
For the shape, weight, density, and gradation of bridge pier riprap, specifications developed for revetment riprap are applicable (Lagasse et al. 2006). These specifications are provided in Design Guideline 4 of this document (see Section 4.2.4).
Design Guideline 4 recommends gradations for ten standard classes of riprap based on the median particle diameter d50 as determined by the dimension of the intermediate ("B") axis. These gradations were developed under NCHRP Project 24-23, "Riprap Design Criteria, Recommended Specifications, and Quality Control." The proposed gradation criteria are based on a nominal or "target" d50 and a uniformity ratio d85/d15 that results in riprap that is well graded. The target uniformity ratio is 2.0 and the allowable range is from 1.5 to 2.5 (Lagasse et al. 2006).
Standard test methods relating to material type, characteristics, and testing of rock and aggregates recommended for revetment riprap are applicable to bridge pier riprap (see Design Guideline 4). In general, the test methods recommended are intended to ensure that the stone is dense and durable, and will not degrade significantly over time.
Rocks used for riprap should only break with difficulty, have no earthy odor, no closely spaced discontinuities (joints or bedding planes), and should not absorb water easily. Rocks comprised of appreciable amounts of clay, such as shales, mudstones, and claystones, are never acceptable for use as riprap. The recommended tests and allowable values for rock and aggregate are summarized in Table 4.3 of Design Guideline 4.
Laboratory studies of bridge pier riprap have been conducted to develop guidance on more than just the design size of bridge pier riprap. These studies have confirmed that a properly designed riprap system must integrate appropriately sized stone with adequate layout dimensions (extent and thickness) and an underlying filter (granular or geotextile) layer. This section summarizes the results of laboratory studies on bridge pier riprap conducted under NCHRP Projects 24-07(1) and 24-07(2) conducted at St. Anthony Falls Hydraulics Laboratory (Parker et al. 1998) and Colorado State University (Lagasse et al. 2007), respectively.
Some studies suggest that a filter may be unnecessary if the riprap layer is of sufficient thickness (Toro-Escobar 1998). Yet, a majority of the research on the stability of riprap at bridge piers to date indicates that the use of an underlying filter layer significantly increases the stability of the riprap layer. Many of the more recent experimental studies have evaluated the effects of a filter layer placed below a riprap layer on the stability of the riprap layer under live-bed conditions.
In general, granular filter layers should be of a gradation, size, and thickness sufficient to deter the effects of winnowing of the underlying bed sediments. Geotextiles should also have an effective pore size sufficiently small to block the passage of bed sediments, but have large enough permeability to deter or withstand buoyant forces and potential pressure gradients in the surface and subsurface in the area of the pier.
Parker et al. (1998) determined that placing a geotextile under a riprap with the same areal coverage as the riprap layer resulted in relatively poor performance of the riprap at bridge piers. As a result of the effects of live-bed conditions described above, the riprap at the edges tended to roll, slide or be plucked off exposing the underlying geotextile and ultimately resulting in failure of the riprap layer as successive bed forms pass and pluck more stones from the riprap layer. The failure of the geotextile was due in part to the impermeability of the fabric leading to the buildup of uplift forces and the creation of a bulge under the fabric, which contributed to the loss of riprap stones. In addition, the loss of the edge riprap and exposure of the geotextile allowed the geotextile to fold back on itself further reducing the stability of the riprap. If the geotextile was not sealed to the pier face, winnowing around the pier face resulted in a scour hole around the pier face and caused the geotextile and stones at the interface to fall into the scour hole.
For bridge piers, Parker et al. (1998) determined that the tendency for riprap to settle was arrested when: (a) the geotextile has 2/3 the areal coverage of the riprap, (b) the geotextile is sufficiently permeable, and (c) the geotextile is sealed to the pier. Lauchlan (1999) recommends that the geotextile have an areal coverage of 75% of the riprap layer so that the edges of the geotextile will be anchored when the edge stone of the riprap layer slide into the trough of passing bed forms.
At Colorado State University (CSU), a matrix of flume tests was completed for the NCHRP 24-07(2) research program (Lagasse et al. 2007). Both clear-water and live-bed conditions were examined. The laboratory tests were not designed to replicate any particular prototype scale conditions. However, in each case, the test countermeasure was "designed" to withstand the 2Vcrit hydraulic condition. For example, the riprap size was selected such that particle dislodgement or entrainment was not anticipated during the 2Vcrit run. This did not mean that the riprap wouldn't fail due to other factors, such as settling, edge undermining, or winnowing of substrate material. Runs utilizing an approach velocity of 2.5Vcrit were intended to take the riprap system to failure by particle dislodgement.
As a baseline, maximum scour was determined for unprotected square and rectangular piers, under clear-water and live-bed conditions. A live-bed test was run for a sufficient duration (8 hours) to permit bed forms to migrate through the system. Figure 11.3 shows the results of unprotected square pier tests under live-bed conditions in the CSU indoor flume. Figure 11.4 shows the results of riprap tests under clear-water (Figure 11.4a) and live-bed (Figure 11.4b) conditions. These tests validated the use of Equation 11.1 for sizing bridge pier riprap and the HEC-23 recommendations for riprap extent.
Typically, riprap used for pier scour protection is placed on the surface of the channel bed, in a pre-existing scour hole, or in a hole excavated around the pier. The Federal Highway Administration (Richardson and Davis 2001, Lagasse et al. 2001) recommends placing the top of the riprap layer flush with the channel bed for inspection purposes.
The design intent for the NCHRP 24-07(2) riprap coverage tests included (Lagasse et al. 2007):
Test results indicated that best performance was achieved when riprap extended at least 2 times the width of the pier (as measured perpendicular to the approach flow on all sides) in a flat pre-excavated hole with the top surface flush with the bed. Figure 11.5 shows the poor performance when the areal coverage was reduced to less than two pier widths on all sides.
Riprap used for pier protection is often placed on the surface of the channel bed because of the ease and lower cost of placement and because it is more easily inspected. Test results indicated that when the stable baseline riprap configuration was mounded on the surface without a filter performance was poor. None of the tests with mounded riprap performed as well as tests with the top of the riprap level with the bed, given the same areal extent of riprap coverage. Figure 11.6 shows the results of a mounded riprap test.
Numerous riprap studies (see Lagasse et al. 2006) suggest that thickness of the riprap layer placed around the bridge piers should be between 2 to 3 times median stone size (2-3d50) of the riprap. Testing results indicate that 3d50 is appropriate for specifying minimum thickness and that performance improved with increasing riprap layer thickness.
As noted, NCHRP Project 24-07(1) (Parker et al. 1998) determined that placing a geotextile under a riprap layer with the same areal coverage as the riprap layer resulted in a relatively poor performance of the riprap. Parker et al. suggested extending the geotextile from the pier to about 2/3 of the way to the periphery of the riprap would result in better performance. Additional test results for NCHRP Project 24-07(2) confirmed that riprap performance was best when a geotextile filter extended 2/3 the distance to the periphery of the riprap (Lagasse et al. 2007).
It was found that granular filters performed poorly in the case where bed forms are present. Specifically, during the passage of dune troughs past the pier that are deeper than the riprap armor; the underlying finer particles of a granular filter are rapidly swept away. The result is that the entire installation became progressively destabilized beginning at the periphery and working toward the pier. Figure 11.7 shows two piers after testing, one pier had a geotextile filter that extended 2/3 the distance from the pier face to the periphery (Figure 11.7a) and the other pier had a granular filter that extended the full distance from the pier face to the periphery of the riprap (Figure 11.7b).
For NCHRP Project 24-07(2) the use of sand filled geotextile containers as a filter under riprap was tested at a prototype scale pier (Lagasse et al. 2007). A test section was created that was 30.7 ft (9 m) long and spanned the width of the flume. It was filled with sand level with the approach section. Upstream and downstream of the test section the flume bed consists of smooth concrete floors. A rectangular pier measuring 1.5 ft (0.5 m) by 4.5 ft (1.5 m)was installed in the center of the test section. Figure 11.8 is a layout diagram for the prototype testing program. Surrounding the pier, a scour hole measuring 12 ft by 16 ft (4m x 5 m) was pre-formed into the sand bed to a maximum depth of 3 ft (0.4 m) as shown in Figure 11.9. For the geotextile containers the test at prototype scale was, primarily, to demonstrate constructability and performance in high velocity flow conditions.
Sand filled geotextile containers were constructed using a geotextile fabric with the characteristics presented in Table 11.1. The geotextile containers measured 4 ft x 1.5 ft x 0.33 ft (1.2 m x 0.5 m x 0.1 m) with a typical volume of 2 ft--3 (0.6 m3). Approximately 220 lbs (100 kg) of sand was placed in each bag. Commercial concrete sand meeting appropriate filter criteria was used to fill the geotextile container. Figure 11.10 shows the geotextile containers before being placed around the pier.
An approach flow 1 ft (0.305 m) deep at approximately 1.5 ft/s (0.5 m/s) was established. A total of 32 geotextile containers were placed around the pier by dropping from a height of about 5 ft (1.5 m) above the water surface. Installation was facilitated by a backhoe fitted with a special grapple attached to the bucket, which enabled the backhoe to pick up the geotextile container, position around the pier to a specified location, and release the container. Figure 11.11 is a photograph of a geotextile container being dropped near the pier; note the grapple plate attachment to the backhoe. Figure 11.12 shows the geotextile containers after installation in approximately 1 ft (0.305 m) of flowing water.
Next, riprap was positioned on top of the geotextile containers using the backhoe with the grapple removed. Figure 11.13 shows riprap being dropped near the pier and Figure 11.14 shows the riprap after installation. These tests confirmed that geotextile containers can be fabricated locally and that the containers and riprap can be placed with standard commercially available equipment. The final step in the testing procedure was to use partial grouting techniques (see Design Guideline 12) to demonstrate the enhanced stability of partially grouted riprap as a pier scour countermeasure.
Based on information derived primarily from NCHRP Project 24-07(2) the optimum performance of riprap as a pier scour countermeasure was obtained when the riprap extended a distance of 2 times the pier width in all directions around the pier (Lagasse et al. 2007).
In the case of wall piers or pile bents consisting of multiple columns where the axis of the structure is skewed to the flow direction, the lateral extent of the protection should be increased in proportion to the additional scour potential caused by the skew. While there is no definitive guidance for pier scour countermeasures, it is recommended that the extent of the armor layer should be multiplied by a factor Kα, which is a function of the width (a) and length (L) of the pier (or pile bents) and the skew angle α as given below (Richardson and Davis 2001):
Riprap should be placed in a pre-excavated hole around the pier so that the top of the riprap layer is level with the ambient channel bed elevation. Placing the top of the riprap flush with the bed is ideal for inspection purposes, and does not create any added obstruction to the flow. Mounding riprap around a pier is not acceptable for design in most cases, because it obstructs flow, captures debris, and increases scour at the periphery of the installation (see Figure 11.6).
The riprap layer should have a minimum thickness of 3 times the d50 size of the rock. However, when contraction scour through the bridge opening exceeds 3d50, the thickness of the riprap must be increased to the full depth of the contraction scour plus any long-term degradation. In river systems where dune bed forms are present during flood flows, the depth of the trough below the ambient bed elevation should be estimated using the methods of Karim (1999) and/or van Rijn (1984). In general, an upper limit on the crest-to-trough height Δ is provided by Bennett (1997) as Δ < 0.4y where y is the depth of flow. This suggests that the maximum depth of the bed form trough below ambient bed elevation will not exceed 0.2 times the depth of flow. Additional riprap thickness due to any of these conditions may warrant an increase in the extent of riprap away from the pier faces, such that riprap launching at a 1V:2H slope under water can be accommodated. When placement of the riprap must occur under water, the thickness should be increased by 50%. Recommended layout dimensions for bridge pier riprap are provided in Figure 11.15.
The importance of the filter component of any riprap installation should not be underestimated. There are two kinds of filters used in conjunction with riprap; granular filters and geotextile filters. Some situations call for a composite filter consisting of both a granular layer and a geotextile. The specific characteristics of the base soil determine the need for, and design considerations of the filter layer. In cases where dune-type bed forms may be present, it is strongly recommended that only a geotextile filter be considered. Guidance on the design of granular and geotextile filters is provided in Design Guideline 16.
A filter layer is typically required for riprap at bridge piers. The filter should not be extended fully beneath the riprap; instead, it should be terminated 2/3 of the distance from the pier to the edge of the riprap. When using a granular stone filter, the layer should have a minimum thickness of 4 times the d50 of the filter stone or 6 in. (15 cm), whichever is greater. As with riprap, the layer thickness should be increased by 50% when placing under water.
Sand-filled geotextile containers made of properly-selected materials provide a convenient method for controlled placement of a filter in flowing water. This method can also be used to partially fill an existing scour hole when placement must occur under water, as illustrated in Figure 11.16. For more detail, see Sections 11.5.3 and 11.6.2.
Placing geotextiles under water is problematic for a number of reasons. Most geotextiles that are used as filters beneath riprap are made of polyethylene or polypropylene. These materials have specific gravities ranging from 0.90 to 0.96, meaning that they will float unless weighted down or otherwise anchored to the subgrade prior to placement of the riprap (Koerner 1998). In addition, unless the work area is isolated from river currents by a cofferdam, flow velocities greater than about 1.0 ft/s (0.3 m/s) create large forces on the geotextile. These forces cause the geotextile to act like a sail, often resulting in wavelike undulations of the fabric (a condition that contractors refer to as "galloping") that are extremely difficult to control. In mild currents, geotextiles (precut to length) have been placed using a roller assembly, with sandbags to hold the fabric temporarily.
To overcome these problems, engineers in Germany have developed a product known as SandMatTM. This blanket-like product consists of two non-woven geotextiles (or a woven and a non-woven) with sand in between. The layers are stitch-bonded or sewn together to form a heavy, filtering geocomposite. The composite blanket exhibits an overall specific gravity ranging from approximately 1.5 to 2.0, so it sinks readily.
According to Heibaum (2002), this composite geotextile has sufficient stability to be handled even when loaded by currents up to approximately 3.3 ft/s (1 m/s). At the geotextile - subsoil interface, a non-woven fabric should be used because of the higher angle of friction compared to woven geotextiles. Figure 11.17 shows a close-up photo of the SandMatTM material. Figure 11.18 shows the SandMatTM blanket being rolled out using conventional geotextile placement equipment.
In deep water or in currents greater than 3.3 ft/s (1 m/s), German practice calls for the use of sand-filled geotextile containers. For specific project conditions, geotextile containers can be chosen that combine the resistance against hydraulic loads with the filtration capacity demanded by the application. Geotextile containers have proven to give sufficient stability against erosive forces in many applications, including wave-attack environments. The size of the geotextile container must be chosen such that the expected hydraulic load will not transport the container during placement (Heibaum 2002). Once placed, the geotextile containers are overlaid with the final armoring material (see Figure 11.16 and Section 11.5.3).
Figure 11.19 shows a large geotextile container being filled with sand. Figure 11.20 shows the sand-filled geotextile container being handled with an articulated-arm clam grapple. The filled geotextile container in the photograph is a nominal 1-metric-ton (1,000 kg or 2,200 lb) unit. The preferred geotextile for these applications is always a non-woven needle punched fabric, with a minimum mass per unit area of 500 grams per square meter. Smaller geotextile containers can be fabricated and handled by one or two people for smaller-sized applications.
Riprap is to be sized for an existing 2 ft (0.61 m) square pier (see Figure 11.21). The average velocity in the channel is 3.8 ft/s (1.16 m/s) and the pier is located in the main current around a mild bend. The average depth of flow is 6 ft (1.8 m). The riprap specific gravity is 2.5. The computed contraction scour is 2.0 ft (0.61 m). No long-term degradation is anticipated at this site.
Step 1: Select the appropriate shape coefficient (K1) = 1.7.
Step 2: Determine the appropriate design velocity:
= (1.7) (1.4) (3.8)
= 9 ft/sec (2.7 m/s)
Step 3: Determine d50 from Equation 11.1:
Step 4: Select Class II riprap from Table 4.1 of Design Guide 4: d50 = 9 in. (0.23 m)
Step 5: Determine the depth of riprap below the streambed at the pier:
The depth of riprap is the greater of 3d50, the contraction scour and long-term degradation depth, or the depth of bedform troughs.
3 d50 = 3 (9) = 27 in. = 2.25 ft (0.69 m)
Contraction Scour = 2.0 ft (0.01 m)
Bed forms = 0.2 (6) = 1.2 ft (0.37 m)
Step 6: Determine the riprap extent:
The recommended extent is at least two times the pier width. Therefore, the minimum riprap extent is 4 ft (1.22 m) from each face of the pier.
Step 7: Additional considerations:
Bennett, J.P., 1997, "Resistance, Sediment Transport, and Bedform Geometry Relationships in Sand-Bed Channels," in: Proceedings of the U.S. Geological Survey (USGS) Sediment Workshop, February 4-7.
Heibaum, M.H., 2002, "Geotechnical Parameters of Scouring and Scour Counter-measures,"Mitteilungsblatt der Bundesanstalt für Wasserbau Nr. 85.(J. Federal Waterways Engineering and Research Institute, No. 85), Karlsruhe, Germany.
Isbash, S.V., 1936, "Construction of Dams by Depositing Rock in Running Water," Transactions, Second Congress on Large Dams, U.S. Government Report No. 3, Washington D.C.
Karim, F., 1999, "Bed-Form Geometry in Sand-Bed Flows," Journal of Hydraulic Engineering, Vol. 125, No. 12, December.
Koerner, R.M., 1998, "Designing with Geosynthetics," 4th Edition, Prentice-Hall, Inc., Englewood Cliffs, NJ, 761 p.
Lagasse, P.F., Zevenbergen, L.W., Schall, J.D., and Clopper, P.E., 2001, "Bridge Scour and Stream Instability Countermeasures," Hydraulic Engineering Circular No. 23 (HEC-23, Second Edition), Report FHWA NHI -01-003, Federal Highway Administration, Washington, D.C.
Lagasse, P.F., Clopper, P.E., Zevenbergen, L.W., and Ruff, J.F., 2006, "Riprap Design Criteria, Recommended Specifications, and Quality Control," NCHRP Report 568, Transportation Research Board, National Research Council, Washington, D.C.
Lagasse, P.F., Clopper, P.E., Zevenbergen, L.W., and Girard, L.G., 2007, "Countermeasures to Protect Bridge Piers from Scour," NCHRP Report 593, Transportation Research Board, National Research Council, Washington, D.C.
Lauchlan, C.S., 1999, "Pier Scour Countermeasures," Ph.D. Thesis, University of Auckland, Auckland, NZ.
Melville, B.W. and Coleman, S.E., 2000, "Bridge Scour," Water Resources Publications, LLC, Highlands Ranch, CO, 550 pp.
Parker, G., Toro-Escobar, C., and Voigt, R.L. Jr., 1998, "Countermeasures to Protect Bridge Piers from Scour," Users Guide (revised 1999) and Final Report, NCHRP Project 24-07, prepared for Transportation Research Board by St. Anthony Falls Laboratory, University of Minnesota, MN, 360 pp.
Richardson, E.V. and S.R. Davis, 2001, "Evaluating Scour at Bridges," Hydraulic Engineering Circular 18, Fourth Edition, FHWA NHI 01-001, Federal Highway Administration, U.S. Department of Transportation, Washington, D.C.
Toro-Escobar, C., Voigt, R., Jr., Melville, B., Chiew, M., and Parker, G., 1998, "Riprap Performance at Bridge Piers Under Mobile-Bed Conditions," Transportation Research Board, Transportation Research Record 1647, Paper No. 98-1165, pp. 27-33.
van Rijn, L.C., 1984, "Sediment Transport, Part III: Bed Forms and Alluvial Roughness," Journal of Hydraulic Engineering, Vol. 110, No. 12, December.